The present invention is directed to a system for finding a transformation of a gray-level pattern in an image. The system recursively searches a transformation space in a multi-resolution manner. At each resolution, the transformation space is divided into groups of transformations. For each group, pixels in the pattern are compared to neighborhoods of pixels in the image in order to identify whether a specific pattern pixel is within the minimum and maximum values of a neighborhood of pixels.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method for determining a transformation of a gray level pattern in an image, comprising the steps of: dividing a set of possible transformations into groups of transformations, each group having a reference transformation, each transformation of said set of possible transformations represents movement of said pattern; determining whether a set of pixel intensities of said pattern are within minimum values and maximum values, said minimum values represent minimum pixel intensities for neighborhoods of pixels in said image and said maximum values represent maxim pixel intensities for neighborhoods of pixels in said image, said step of determining being performed for each reference transformation; applying a difference function based on said step of determining, said step of applying being performed for each reference transformation; and identifying a particular transformation which meets a predefined criteria with respect to said difference function.
2. A method according to claim 1, further including the step of: discarding transformations in groups of transformations having reference transformations with a difference function worse than a previously determined best known difference value.
3. A method according to claim 1, further including the steps of: continuously dividing a transformation space into smaller groups until a minimum group size is considered, each of said groups belonging to one level of a set of levels, said step of dividing a transformation space includes choosing at least one group for dividing in each of at least a subset of said levels, said step of choosing includes said steps of determining whether a set of pixel intensities, applying a difference function and identifying a transformation; and backtracking through said levels including removing groups from consideration having a difference value worse than a best known difference value and performing said steps of continuously dividing on said groups having a difference value better than said best known difference value, subsequent to finally completing said step of backtracking said best known difference value corresponds to said transformation of said gray level pattern in said image.
4. A method according to claim 3, wherein: said groups are two dimensional.
5. A method according to claim 3, wherein: said groups are six dimensional.
6. A method according to claim 3, further including the steps of: receiving said pattern; and receiving said image.
7. A method according to claim 3, further including the step of: reporting said transformation of said gray level pattern in said image.
8. A method according to claim 3, wherein: groups for a first level have a first groups size; and neighborhoods for said first level have a second size, said second size being determined by said first size.
9. A method according to claim 3, wherein said step of determining whether a set of pixel intensities of said pattern are within minimum values and maximum values includes the steps of: comparing a pixel intensity of said pattern to a first data value at a transformed location in a map of minimum values; and comparing said pixel intensity of said pattern to a second data value at a transformed location in a map of maximum values.
10. A method according to claim 9, wherein said step of determining whether a set of pixel intensities of said pattern are within minimum values and maximum values further includes the steps of: setting a pixel-region difference as zero if said pixel intensity is less than said second data value and greater than said first data value; setting said pixel-region difference as a difference between said first data value and said pixel intensity if said pixel intensity is less than said first data value; and setting said pixel-region difference as a difference between said pixel intensity and said second data value if said pixel intensity is greater than said second data value.
11. A method according to claim 10, wherein: said step of applying a difference function applies said difference function to said pixel-region difference.
12. A method for determining a transformation of a gray level pattern in an image, comprising the steps of continuously dividing a set of possible transformations into smaller groups until a minimum group size is considered, each of said groups belonging to one level of a set of levels, each transformation of said set of possible transformations represents movement of said pattern, said step of dividing includes choosing at least one group for dividing in each of at least a subset of said levels, said step of choosing includes determining a difference value for a subset of groups based on determining whether a set of pixel intensities of said pattern are within minimum values and maximum values, said minimum values represent minimum pixel intensities for neighborhoods of pixels in said image and said maximum values represent maximum pixel intensities for neighborhoods of pixels in said image; and backtracking through said levels including removing groups from consideration having a difference value worse than a best known difference value and performing said step of continuously dividing on said groups having a difference value better than said best known difference value, subsequent to finally completing said step of backtracking said best known difference value corresponds to said transformation of said gray level pattern in said image.
13. A method according to claim 12, wherein: said step of choosing includes identifying a first set of one or more reference transformations for said pattern; said step of determining a difference value uses a difference function; said step of determining a difference value is performed on at least a subset of said reference transformations; and said step of choosing further includes identifying at least one reference transformation which meets a predefined criteria with respect to said difference function.
14. A method according to claim 12, further including the steps of: receiving said pattern; receiving said image; and reporting said transformation of said gray level pattern in said image.
15. A method according to claim 12, wherein: said step of determining whether a set of pixel intensities of said pattern are within minimum values and maximum values includes the steps of: comparing a pixel intensity of said pattern to a first data value at a transformed location in a map of minimum values, comparing said pixel intensity of said pattern to a second data value at a transformed location in a map of maximum values, setting a pixel-region difference as zero if said pixel intensity is less than said second data value and greater than said first data value, setting said pixel-region difference as a difference between said first data value and said pixel intensity if said pixel intensity is less than said first data value, and setting said pixel-region difference as a difference between said pixel intensity and said second data value if said pixel intensity is greater than said second data value.
16. A method for finding a gray level pattern in an image, comprising the steps of: identifying said pattern in a first image, said pattern being at a first position in said first image; receiving a second image; and determining a transformation of said pattern in said second image based on said first position, said transformation represents movement of said pattern from said first position, said step of determining a transformation includes: identifying a first set of one or more possible reference transformations for said pattern, determining whether a set of pixel intensities of said pattern are within minimum values and maximum values, said minimum values represent minimum pixel intensities for neighborhoods of pixels in said second image and said maximum values represent maximum pixel intensities for neighborhoods of pixels in said second image, said step of determining being performed for each reference transformation in said first set, applying a difference function based on said step of determining, said step of applying being performed for each reference transformation in said first set, and identifying a particular reference transformation which meets a predefined criteria with respect to said difference function.
17. A method according to claim 16, wherein: said pattern is at a second position in said second image; receiving a third image; and determining a transformation of said pattern in said third image based on said second position.
18. A method according to claim 16, further including the step of: reporting a position of said pattern in said second image based on said transformation of said pattern in said second image.
19. A method according to claim 16, wherein: said step of determining whether a set of pixel intensities are within minimum values and maximum values includes the steps of: comparing a pixel intensity of said pattern to a first data value at a transformed location in a map of minimum values, comparing said pixel intensity of said pattern to a second data value at a transformed location in a map of maximum values, setting a pixel-region difference as zero if said pixel intensity is less than said second data value and greater than said first data value, setting said pixel-region difference as a difference between said first data value and said pixel intensity if said pixel intensity is less than said first data value, and setting said pixel-region difference as a difference between said pixel intensity and said second data value if said pixel intensity is greater than said second data value.
20. A method according to claim 19, wherein: said step of applying a difference function, applies said difference function to said pixel-region difference.
21. A processor readable storage medium having processor readable code embodied on said processor readable storage medium, said processor readable code for programming a processor to perform a method for determining a transformation of a gray level pattern in an image, the method comprising the steps of: continuously dividing a set of possible transformations into smaller groups until a minimum group size is considered, each of said groups belonging to one level of a set of levels, each transformation of said set of possible transformations represents movement of said pattern, said step of dividing includes choosing at least one group for dividing in each of at least a subset of said levels, said step of choosing includes determining a difference value for a subset of groups based on determining whether a set of pixel intensities of said pattern are within minimum values and maximum values, said minimum values represent minimum pixel intensities for neighborhoods of pixels in said image and said maximum values represent maximum pixel intensities for neighborhoods of pixels in said image; and backtracking through said levels including removing groups from consideration having a difference value worse than a best known difference value and performing said step of continuously dividing on said groups having a difference value better than said best known difference value, subsequent to finally completing said step of backtracking said best known difference value corresponds to said transformation of said gray level pattern in said image.
22. A processor readable storage medium according to claim 21, wherein: said step of choosing includes the step of identifying a first set of one or more reference transformations for said pattern; said step of determining a difference value uses a difference function; said step of determining a difference value is performed on at least a subset of said reference transformations; and said step of choosing further includes identifying at least one reference transformation which meets a predefined criteria with respect to said difference function.
23. A processor readable storage medium according to claim 21, wherein said method further including the steps of: receiving said pattern; receiving said image; and reporting said transformation of said gray level pattern in said image.
24. A processor readable storage medium according to claim 21, wherein said step of determining whether a set of pixel intensities of said pattern are within minimum values and maximum values includes the steps of: comparing a pixel intensity of said pattern to a first data value at a transformed location in a map of minimum values; and comparing said pixel intensity of said pattern to a second data value at a transformed location in a map of maximum values.
25. A processor readable storage medium according to claim 24, wherein said step of determining whether a set of pixel intensities of said pattern are within minimum values and maximum values includes the steps of: setting a pixel-region difference as zero if said pixel intensity is less than said second data value and greater than said first data value; setting said pixel-region difference as a difference between said first data value and said pixel intensity if said pixel intensity is less than said first data value; and setting said pixel-region difference as a difference between said pixel intensity and said second data value if said pixel intensity is greater than said second data value.
26. An apparatus for determining a transformation of a gray level pattern in an image, comprising: an input device; a display for showing said image; a processing unit in communication with said input device and said display; and a processor readable storage device in communication with said processing unit, said processor readable storage device storing processor readable code, said processor readable code for programming said processing unit to perform a method comprising the steps of continuously dividing a set of possible transformations into smaller groups until a minimum group size is considered, each of said groups belonging to one level of a set of levels, said step of dividing includes choosing at least one group for dividing in each of at least a subset of said levels, each transformation of said set of possible transformations represents movement of said pattern, said step of choosing includes determining a difference value for a subset of groups based on determining whether a set of pixel intensities of said pattern are within minimum values and maximum values, said minimum values represent minimum pixel intensities for neighborhoods of pixels in said image and said maximum values represent maximum pixel intensities for neighborhoods of pixels in said image; and backtracking through said levels including removing groups from consideration having a difference value worse than a best known difference value and performing said step of continuously dividing on said groups having a difference value better than said best known difference value, subsequent to finally completing said step of backtracking said best known difference value corresponds to said transformation of said gray level pattern.
27. An apparatus according to claim 26, wherein: said step of choosing includes the step of identifying a first set of one or more reference transformations for said pattern; said step of determining a difference value uses a difference function; said step of determining a difference value is performed on at least a subset of said reference transformations; and said step of choosing further includes identifying at least one reference transformation which meets a predefined criteria with respect to said difference function.
28. An apparatus according to claim 26, wherein: said step of determining whether a set of pixel intensities of said pattern are within minimum values and maximum values includes the steps of: comparing a pixel intensity of said pattern to a first data value at a transformed location in a map of minimum values, comparing said pixel intensity of said pattern to a second data value at a transformed location in a map of maximum values, setting a pixel-region difference as zero if said pixel intensity is less than or equal to said second data value and greater than or equal to said first data value, setting said pixel-region difference as a difference between said first data value and said pixel intensity if said pixel intensity is less than said first data value, and setting said pixel-region difference as a difference between said pixel intensity and said second data value if said pixel intensity is greater than said second data value.
29. An apparatus according to claim 26, wherein: said input device is a video camera capable of capturing video images; and said method is capable of finding transformations of said pattern in successive video images captured by said video camera.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
June 16, 1998
June 19, 2001
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.